Hey Product Hunt, I build with AI agents every day. I run my whole one-person studio with a couple of them. The thing that kept breaking was memory. Every session started from zero, so I'd re-explain the same context, the same decisions, the same state, over and over. The agents were capable but had the memory of a goldfish. So I built Clawtex to give them an actual brain. It's a structured memory layer: live state (what's true right now), a full event history (what's happened and why), and lessons (what they've learned, so it compounds instead of resetting). It works across every session and surface over MCP, so any Claude instance picks up exactly where the last one left off. It's been running my real studio day to day, two agents sharing one memory, no cold starts. Now it's open for anyone building with Claude. Free to start. I'd genuinely love your feedback, especially from others building agent workflows. What's the one thing you'd want your agents to actually remember? Cheers, Paul

Clawtex

Structured memory for Claude that compounds over time

Structured memory for Claude that compounds over time